By Julian Dierkes
Last year, it was certainly noted in Mongolia that the country had fallen from the status of an “electoral democracy” to an “electoral autocracy”. With 2025, Mongolia is back among the ranks of electoral democracies (see p. 15 of V-Dem’s “Democracy Report 2026” (PDF)).
Cause for celebration?
I’m referring to a post I had published last year that reported on conversations regarding self-censorship that I had in Ulaanbaatar last Spring to make sense of a continued slide in democracy rankings that did not make sense to me given little change in legislative or structural political factor.
To make substantive sense of the uptick, let’s look more closely at subcomponents, i.e. the five different democracy indices that V-Dem provides.
| 2015 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | |
| Deliberative Democracy | 0.56 | 0.48 | 0.45 | 0.43 | 0.41 | 0.4 | 0.45 |
| Egalitarian Democracy | 0.53 | 0.46 | 0.43 | 0.39 | 0.36 | 0.35 | 0.38 |
| Electoral Democracy | 0.68 | 0.63 | 0.61 | 0.57 | 0.53 | 0.5 | 0.54 |
| Liberal Democracy | 0.56 | 0.52 | 0.49 | 0.46 | 0.41 | 0.39 | 0.42 |
| Participatory Democracy | 0.38 | 0.33 | 0.32 | 0.29 | 0.27 | 0.25 | 0.28 |
I have included the values for 2015, and 2020-2025 here. This marks the decline of most values that started in 2019, as well as the opportunity to offer 5- and 10-year comparisons.
10-Year Trend
If we look at the ten year trajectory it is definitely negative. Regardless of the magnitude of year-to-year changes, we thus can concluded that democracy along the five dimensions is less solid now that it was ten years ago. One might note in this context that the MPP has been in government since the 2016 elections and that the DP has been actively self-destructing since around then, despite Pres. Battulga electoral victory in 2017. The only significant democratization in this period was the 2017 initiative toward deliberative polling by current PM Zandanshatar, though this has had no real long-term impact. At the same time, there have been no major crises of democracy, I would argue, on the scare of the 2008 fatal riots, for example. The parliamentary elections during this time (2016, 2020, 2024) were fascinating for many reasons, but I witnessed the 2016 and 2024 election in country, observing the campaign as well as the election, itself, but I am not sure that they were particular markers in terms of democratization. Yes, electoral systems changed every time, but less or more democracy? If yes, then only marginally, I would argue. For example, the creation of regional election districts in 2024 coupled with the short campaign period has surely raised barriers for independent candidates or small parties, but an overall decline in the quality of democracy?
Over this period it is clear, however, that the perception of democratic participation and opportunities has deteriorated.
5-Year Trend
What about the 5-year trend then? Over this period as well, all five subindicators are down. Deliberative democracy least, liberal democracy most (numerically and in percent). The liberal democracy indicator is perhaps easiest to make sense of with its emphasis on civil liberties. As I wrote last year, contacts are increasingly reporting that they are self-censoring. There also have been several incidents of arrests of demonstrators or journalists (for example the noorog.mn case). These indicate a sense of reduced freedoms, especially significant as many Mongolians’ sense of democracy is focused more on freedoms and personal liberty than collective deliberation, perhaps. Deliberative democracy measures inclusion of multiple views in the policy-making process, by contrast, and little has changed in that regard.
Recent Developments
Whenever anyone is politically wronged in Mongolia, they quickly reach for “demise of democracy” arguments. This has been going on for years and we have discussed several such moments on this blog over the years. Is Pres. Khurelsukh’s proposed legislation for the recall of MPs a real threat to democracy? Perhaps not directly, but along with previous developments, perhaps this is death by a thousand cuts?
In short, I continue to think that the decline since 2019 overstated threats to democracy, but this last year’s score also does not capture any real changes. That leaves the question of what’s driving this shift in scores.
Events and Substantive Changes to Democracy
For an explanation, as a Mongolia specialist, I try to think about factors as I did in last year’s post. What might have driven this year’s uptick could be the protests that proceeded the end of the Oyun-Erdene government and the peaceful transition of power from PM Oyun-Erdene to PM Zandanshatar through parliamentary means. For 2026, one might argue that the proposal to enable an easier recall of MPs by Pres Khurelsukh suggests an attempt to consolidate power and punish errant MPs, not a great democratizing initiative.
Methodology
An alternative place to look for an understanding of the changes in scores is methodology which is Kyle Marquardt‘s approach. He is involved in V-Dem at the central level, i.e. as a comparative politics specialists, while my involvement with the index is that of a country expert.
This post raises a several important questions about the @vdeminstitute.bsky.social indices that I’ll try to discuss broadly, then specifically in the case of Mongolia.
Important caveat: I am not a Mongolia expert, so can’t speak substantively about the case; I can discuss the V-Dem methodology.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
One of characteristics that distinguishes V-Dem from some other efforts is methodological care. In many ways, of the multitude of global indices that have sprung up over the past decades, it is the most academic.
Confidence
The first point is that all of the V-Dem democracy indices include a measure of statistical uncertainty in addition to our best estimate of the concept. That is, we provide a range of plausible values for each country-year level of democracy. The shaded area around each line shows this range.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
The inclusion of measures of uncertainty is one of the markers of V-Dem. In coding, I’ve very much appreciated the opportunity to estimate a factor, but to “admit” that I am not as sure of that factor as others that I am very confident on. If we take the subindicator that has declined most, liberal democracy, and compare the 2020 and 2025 values, 0.52 and 0.42, then note that the confidence interval (the shaded area, Marquardt refers to) stretches from 0.47-0.58, for 2025 this ranges from 0.36-0.47. The lowest plausible value within the confidence interval in 2020 includes the highest plausible value in 2025. Yet, the consistent trend over time is also suggestive, even given confidence ranges.
These credible intervals are important bc they reveal in which country-years we’re more or less confident about a value, with wider intervals indicating more uncertainty. More uncertainty can be due to multiple factors, from expert uncertainty to different values for index components…
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
Without listing the values individually, the ranges for the subindicators have remained stable over the 5-year period.
They’re also important because they help determine if a change is substantial or not. For example, how confident are we that electoral democracy in Mongolia increased btw 2024 and 2025?
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
Since the confidence bands stayed relatively stable across the indicators over this period, it does not seem to me like there are significant differences in confidence of annual score changes. The confidence in the decline of 2023-2024 is thus not particular greater or lesser than that in the uptick 2024-2025.
In 2024, the level of electoral democracy in Mongolia was .50, with a credible interval of .44 to .56; this indicates that we’re 68% certain the value of electoral democracy was within this range. In 2025, the equivalent statistics were .54 (.47, .60).
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
This is a more precise version of what I commented on above in terms of the confidence bands.
As a rough heuristic, the substantial overlap in the two ranges (.44, .56) and (.47, .60) indicates that we have a relatively low level of confidence that there was a meaningful increase in the level of electoral democracy in Mongolia between 2024 and 2025.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
I find this interpretation quite telling, particularly since Marquardt is directly involved in V-Dem and thus knows the data very well. This also reinforces my own sense that I am much more interested 5-year or 10-year trends when it comes to a single case, Mongolia, than I am in annual changes, though media reporting begs comment on the annual changes.
On the other hand, in 2015 the level of electoral democracy in Mongolia was .69 (.61, .75), which does not overlap with the range in 2024 (.44, .56). This lack of overlap indicates that we can be confident there was a meaningful decrease over that period.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
Bingo! The ten year trend!
Specific Trends
The second point, as Dan mentioned in another thread, is that V-Dem indices include a lot of information about different concepts. The use of indices is useful for understanding broad trends, but not as helpful for understanding the specifics of these trends.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
The general point here, “useful for broad trends, less so for specifics” is similar to my conclusion that I prefer 5/10-year trends. I would say, however, that as a country specialist I am interested in specifics of these trends. It is an understanding of these specifics that enables use of these indices beyond the important academic variable they provide for crossnational comparison, i.e. benchmarking or accountability. If I take dedication to democracy among Mongolian officials at face value (I have little reason not to), then I might imagine that they might be interested in how Mongolia measures up.
To understand the specifics of the trends, it’s useful to drill-down into specific indices. For example, all V-Dem democracy indices include electoral democracy as an integral component. As a result, similar trends across these indices could just be due to electoral democracy decreasing…
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
I would believe that there is a bit of “pollution” of different variables across the different components that make up these indices. If something really terrific happens on a particular issue (imagine some significant victory in a freedom of speech case, for example), it seems plausible that experts will be in a more positive “mood” regarding democracy overall. There are lots of factors that guard against that in the V-Dem methodology (precise focus on particular questions, vignettes to assess an overall general understanding of a topic/question, etc.). In correspondence with Staffan Lindberg who is Director of the V-Dem Project at the Univ of Gothenburg, last year, I learned about the “bad vibes bias” as a technical term for what I am describing here. He pointed me to “Conceptual and Measurement Issues in Assessing Democratic Backsliding” as a detailed examination and discussion of the risks of susceptibility to biases. Both, Marquart and Lindberg were co-authors of that analysis. For anyone who is interested in a very thorough analysis of biases in the V-Dem data, this is a fascinating read. Importantly, the analyses across V-Dem scoring find no evidence for a bad vibes bias, or the claim that “experts have been increasingly negatively biased due to pessimism about the state of democracy” (in the above article).
The evidence across all scores seems quite convincing, yet, my close focus on a single case still has me wondering…
Here I’ve disaggregated the democracy indices into their component indices (i.e. the unique aspect of each democracy index, and electoral democracy on its own). They all seem to show a similar pattern 2015-2025: a decrease btw 2015 and 2024, followed by an uptick in 2025. However…
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
…all of the components are marked by high uncertainty, and the rate and extent of change varies substantially. Electoral democracy shows the greatest degree of change btw 2015 and 2024, which could further explain some of the similarity in changes in all the indices in this period.
— Kyle Marquardt (@kailmarkvart.bsky.social) March 19, 2026 at 9:28 AM
It is in part this disaggregation that makes me wonder. Following Mongolian events closely, I know that my own scores for V-Dem have not changed much, because I do not see a lot of concrete changes, mostly “bad vibes”. As I discussed in my post on self-censorship, those bad vibes may well be adding up to changes in the practice of democracy by not actually restricting free speech (the kind of legislative change that would be privileged in other indices like Freedom in the World, for example), but rather not only the practice of free speech by potential political speakers, but also the perception of the state of freedom of speech. Even if that is the case, these bad vibes may well be dragging down the scores on other specific elements in the disaggregation that actually have little to do with free speech, for example. I am less concerned with global bad vibes in this case, more concerned about bad vibes within the crew of Mongolia experts, in part due to our/their narrow focus on a single case which can make it difficult to make judgements in a comparative context.
I tried to reflect on this already years ago in acknowledging my perhaps somewhat overly-optimistic views of Mongolian democracy. The challenge of providing a thorough assessment of the current state of affairs in Mongolia while also acknowledging that democracy remains fundamentally sound (in my judgement), particularly in comparison to many other jurisdictions remains real. I do think that even when I feel compelled to comment on annual changes because these appear in the media, a strict analytical focus on longer-term developments guards against that somewhat.
